Agent identity for the web
Agents need stable, accountable identities instead of pretending to be humans inside brittle browser flows.
The old pattern is too fragile
Most agents on the web still inherit systems that were designed for one person sitting behind one browser. They reuse cookies, drive interfaces made for humans, and work around detection systems that cannot tell the difference between abuse and delegated work.
That can work for demos, but it is a shaky foundation for real economic activity. When an agent registers accounts, receives verification codes, or sends messages, the service on the other side needs to know more than whether the traffic looks human.
Identity should travel with delegated work
The more useful question is not whether an actor is human. It is whether the actor is accountable, authorized, and attached to a durable root identity. An agent should be able to present itself as software while still proving that a real user delegated the work.
Stable identifiers make that possible. They give developers a way to scope permissions, set limits, audit behavior, and recover from abuse without forcing every agent to masquerade as a person clicking through a page.
Phone numbers are a practical starting point
Phone numbers are not a complete identity system, but they are already woven into account creation, recovery, messaging, and verification. They create a bridge between software agents and the services that still depend on familiar trust signals.
OP starts there: real phone numbers exposed through APIs so agents can operate with a persistent identity that is explicit, programmable, and easier for services to reason about.
For a practical phone identity example, see creating a Google account with a browser agent and an OP Number. For the phone-number reputation piece, read VoIP numbers vs carrier numbers for agents.